| |
|---|---|
 (630Kb)
|
| Haruo Sawada | |
| Head | |
| World Forest Monitoring Research Team | |
| Forestry and Forest products Research Institute, Japan | |
| Ministry of Agriculture, Forestry and Fisheries | |
| 1 Matsunosato, Kukizaki-machi, Ibaraki, 305-8687, Japan | |
| Tel: +81-298-73-3211 Fax: +81-298-73-1541 | |
| E-mail: sawady@ffpri.affrc.go.jp |
11,445,000 Yen
(FY 1999: 3,828,000 Yen)
Recently, improvement of forest survey and fusing between ground survey information and satellite information have been required, since it is correspondent to global environmental problem and also to forest management problem. In the multistage observation, the pixel decomposition method, which can estimate the component in the pixel of satellite data for the rough resolution, seems to be very much the importance. Then, the elucidation of characteristics of the pixel decomposition method is required. The analysis which adapted to the forest observation asks the requirements, such as 1) the situation in a pixel can be analogized, 2) the effect of the landform is small, 3) forest stand structure such as the evergreen/deciduous mixture rate of the upper layer tree can be distinguished, 4) the change of forest stand structure is proven even in the condition in which the vegetation is comparatively dense.
At first, the basic patterns were decided for applying the pattern decomposition method to forest area. Then, the pattern decomposition method was processed at each pixel, and the relevant analysis between the decomposition coefficient and forest survey data was carried out. As the result, the followings were proven; (1) By pattern decomposition it is possible to obtain information on the forest structure. (2) The conifer pattern decomposition coefficient (CPI) is proportional to the crown covering of the evergreen needle-leaved tree. (3) Broad-leaved tree pattern decomposition coefficient (VPI) reacts at the total amount of the vegetation. (4) The soil pattern decomposition coefficient (SPI) reacts to deciduous leaf quantity and soil. (5) The effect of the atmosphere to the VPI is less than that to the CPI. (6) The effect of the landform is reduced by the pattern decomposition coefficient. (7) In the Aomori cypress primeval forest zone, the effectiveness of pattern decomposition method could be confirmed. (8)The effectiveness of pattern decomposition method was confirmed even in the Kiso hinoki primeval forest zone. (9) Qualitative distribution information of coniferous trees with 30-meter resolution in the district forestry office unit was obtained for the first time.
remote sensing, pattern decomposition, forest, mixel